-----------------------------------------------------------------------Copyright © 1998 by The Johns Hopkins University Press. All rights reserved. Kennedy Institute of Ethics Journal 8.2 (1998) 131-144 ------------------------------------------------------------------------ Access provided by your local institution The Question Not Asked: The Challenge of Pleiotropic Genetic Tests Robert Wachbroit ------------------------------------------------------------------------ Abstract. Nearly all of the literature on the ethical, legal, or social issues surrounding genetic tests has proceeded on the assumption that any particular test for a gene mutation yields information about only one disease condition. Even though the phenomenon of pleiotropy, where a single gene has multiple, apparently unrelated phenotypic effects, is widely recognized in genetics, it has not had much significance for genetic testing until recently. In this article, I examine a moral dilemma created by one sort of pleiotropic testing, APOE genotyping, which can yield information about the risk of two different conditions--coronary heart disease and Alzheimer's disease. A physician administering APOE testing for the beneficial purpose of assessing the risk of heart disease may discover medically useless and socially harmful information about the patient's risk of Alzheimer's disease. I explore how much providers should disclose to patients about pleiotropic test results and whether patients are obligated to know as much about their genetic condition as possible. Discussion regarding genetic testing have gone through two stages. Inthe first, the focus was primarily on single- gene disorders--those in which a particular gene mutation is causally necessary and sufficient for a disease to occur. These discussions centered on the ability of genetic tests to predict the onset of a disease or to confirm its diagnosis and on the associated ethical issues, such as who should have access to this information, what restrictions should be placed on its use, and, in the case of diseases where therapies were not yet available, whether the genetic tests should be performed at all. While these concerns were important, they did not seem applicable to many situations, since single-gene disorders such as Huntington's disease are relatively rare. [End Page 131] In the second stage, attention has centered on more common diseases to which genes have a more complex relation--those for which a genetic mutation may be neither causally necessary nor sufficient for the occurrence of the disease. Discussions now focused on diseases such as cystic fibrosis, various types of cancer, and heart diseases, where the presence of particular environmental conditions or family histories is an essential part of the causal nexus that links a specific genetic mutation and a disease. (Indeed, in some cases the genetic mutation may not even be a partial cause but rather a risk factor--e.g., the genetic mutation may be associated with a disease via a causal factor common to them both.) These discussions have responded to the complexity of the connection between genes and diseases by adding considerations of risk communication and risk perception to the earlier concerns about privacy, discrimination, and other issues. For the most part, second-stage discussions have addressed only one type of complex causal relation, called "heterogeneity," where multiple gene mutations are involved in a single disease. The converse type of complexity, called "pleiotropy," where a single-gene mutation is involved in multiple, apparently unrelated diseases, has received hardly any attention at all. This relative neglect, insofar as it is intentional, might indicate a belief that pleiotropy raises no special ethical issues. Perhaps it has been assumed that if a single genetic test indicates a risk of contracting two (apparently unrelated) diseases, the ethical issues it raises are identical to those raised if two distinct tests, one for disease A and the other for disease B, had been performed. On this view, pleiotropic genetic tests may be more efficient ("two conditions for the price of one"), but they do not raise ethical questions different from those created by two separate tests. This conclusion would be justified if the response to concerns raised by disease A were always compatible with the response to concerns raised by disease B. If, however, there is a conflict--e.g., the recommendations regarding testing for disease A conflict with the recommendations regarding testing for disease B--then the pleiotropic genetic test would appear to present a special ethical issue. The latter possibility is not merely theoretical. The test for the allele of the APOE gene is a case in point. Apoe Genotyping Coronary heart disease (CAD) and Alzheimer's disease (AD) are two apparently unrelated diseases. Nevertheless, testing for the *4 allele of the apolipoprotein E (APOE) gene can yield information about both conditions.[End Page 132] A positive test result indicates a high risk of hyperlipidemia and, consequently, a high risk of atherosclerosis and myocardial infarction (Wilson et al. 1994). At the same time, for reasons we do not fully understand, a positive test result is strongly associated with a high risk of contracting AD (Strittmatter et al. 1993). Because of the profound differences in our understanding of these two diseases and in their social meanings, this pleiotropic test appears to generate a moral dilemma. On the one hand, since CAD is in many cases treatable, there are clear medical benefits to administering an APOE test to identify those at risk. Problems of discrimination in employment or insurance as a result of being tested might arise in particular situations, but they rarely will outweigh the benefits of treatment. Thus the APOE test is part of the cardiologist's standard repertoire. On the other hand, since AD is, as yet, untreatable, the medical benefit of a test that might identify those at risk is unclear. Furthermore, there are considerable uncertainties in the interpretation of an APOE test for AD. Because of these uncertainties as well as the lack of a treatment, many researchers have concluded that APOE genotyping should not be used either as the sole diagnostic tool or as a predictive tool for AD (Mayeux et al. 1998). Given the various harms resulting from the belief that an individual is at risk for AD, including stigmatization, discrimination, and the psychological burden associated with the prospect of having AD, several commentators and organizations, including the National Institute on Aging, the Alzheimer's Association (NIA/AA 1996), the American College of Medical Genetics, and the American Society of Human Genetics Working Group on ApoE and Alzheimer Disease (ACMG/ASHG 1995), have recommended against using the APOE test at this time to assess AD risk. Here we have one of the starkest conflicts generated by a pleiotropic genetic test. In the context of CAD, APOE genotyping is clinically useful and accordingly recommended when appropriate. In the context of AD, however, APOE genotyping is deemed harmful and is not recommended. Thus, we can have a situation in which a physician administers APOE testing for the beneficial purpose of CAD risk assessment but, in the process, discovers socially harmful or medically useless information about the patient's risk of AD. In what follows, I will explore how such apparent conflicts might be resolved. I focus on the APOE test in order to make the discussion of pleiotropic genetic testing concrete. The APOE test exemplifies the type of moral dilemma that seems to emerge from pleiotropic genetic testing. It is important [End Page 133] to realize, however, that the particular conflict between the benefits and the harms of APOE testing depends upon the current state of medical science. The availability of a treatment for AD would most likely change the nature of the case but the general questions would remain: When a genetic test may reveal information about two (or more) unrelated conditions, how much should providers disclose to patients? Further, should patients feel obligated to know as much as possible about their genetic condition? As I explore the first question, I shall assume that if the *4 allele of the APOE gene were associated only with AD, testing would not be performed except possibly to support a diagnosis of AD. Later, in an effort to answer the second question, I will drop that assumption (cf. Wachbroit 1996). Is There an Obligation to Disclose? Thus far I have presented the conflict arising from pleiotropic genetic tests only as whether to administer such a test--e.g., do the medical benefits of a CAD risk assessment outweigh the social and psychological costs of an apparent AD risk assessment? However, none of the organizations that have recommended against using APOE genotyping for AD risk have made a recommendation against the clinical use of APOE altogether. No one has urged that cardiologists cease using APOE genotyping for the purpose of CAD risk assessment. Indeed, it would be difficult for health professionals to justify forgoing the clear medical benefits of performing the test simply because of the possible social and psychological costs of learning unintended results. Consequently, the issue is not whether the health professional should ever consider administering the test but rather, when a particular use of the test is indicated and appropriate, what the patient should be told. There are three possible responses: compartmentalization, partial disclosure, and consolidation. Compartmentalization consists of treating the pleiotropic test as if it were two separate tests. With this approach, when the cardiologist uses APOE testing for CAD risk, he or she ignores any information obtained that might indicate a condition other than the one being tested for. By classifying genetic tests in terms of the conditions that they test for, we can easily fail to see the conflicts arising from pleiotropy. For example, in the ACMG's recommendation not to offer APOE testing for susceptibility to AD, no mention is made of the use of this test in identifying CAD risk. It focuses exclusively on one disease. [End Page 134] Part of the rationale for compartmentalization is that it is of a piece with medical specialization. A neurologist, for example, is not likely to order tests regarding a patient's cardiac condition. One could argue that the results of pleiotropic tests are intended as tools for particular specialties, so that, when medical information is revealed that is outside the clinician's specialty, she can regard it as none of her concern. Nevertheless, this appeal to medical specialization cannot take us very far. No one would argue that if a neurologist prescribes medication, he or she can ignore any cardiac effects the drug might have. Similarly, genetic tests can have "side effects"--consequences in terms of yielding information other than that explicitly being sought. Acting as if they do not is indefensible because the responsibilities of physicians towards their patients are not so neatly compartmentalized. The difficulties raised by compartmentalization might suggest the need for a more flexible approach--partial disclosure. This approach consists of not informing the patient of the alternative, and presumably unintended, use of the pleiotropic test before the test is administered, but informing the patient later if the test indicates that the patient is at risk for the other disease. Thus the cardiologist practicing partial disclosure uses the APOE test without telling the patient that it may reveal the risk of AD as well as CAD. The patient is informed of this only if the test reveals an AD risk. This approach can be justified by drawing an analogy to what patients must be told about medication. Only the adverse side effects of a drug need to be disclosed, not all side effects. Similarly, with the partial disclosure approach, only the adverse side effects of a genetic test, should they occur, need to be disclosed. Part of the appeal of this approach is that it does not burden health professionals with matters outside their area of specialization. The cardiologist administering an APOE test can be a cardiologist, confining herself to investigating cardiac issues. She needs to alert the patient to neurological issues only if the test indicates that the patient is at risk for AD. Furthermore, partial disclosure avoids having the benefits of the test from the standpoint of CAD overshadowed by the harms from the standpoint of AD. If patients were told that the test could reveal whether they were at risk for AD, some of them might choose not to be tested. They would rather not know anything about their risk of AD, even if that meant forgoing the benefits of learning about their risks of CAD. Nevertheless, this partial disclosure approach faces at least two immediate problems. First, the analogy to disclosing a drug's adverse side effects [End Page 135] is flawed in several respects. In some situations, especially where close relatives are suffering from a genetic disease, even a negative test result--showing no increased risk of the disease--can be unwelcome. Some patients might experience "survivor guilt" as they contemplate the prospects of their less fortunate siblings or other relatives. The fear of survivor guilt is sometimes an important reason why patients do not want to be tested for a genetic condition. Thus, unlike the physician in the medication analogy, a physician performing a genetic test would have to treat every test result as having a potential adverse side effect. One could also argue that since learning that one is at risk of AD is itself a harm--albeit a social harm or psychological burden--the medication analogy completely fails. The reasoning that leads to informing patients about the risks of adverse side effects from medication before they take the medication should also lead to informing patients about the adverse effects of learning that one is at risk for AD before they take the genetic test. A different way of justifying partial disclosure might be based on problems with the quality of the information about AD risk obtained from the test. One does not have to inform a patient of all possible risks or consequences arising from a particular medical procedure, but only those that are well established. As noted earlier, there is some uncertainty about the predictive value of APOE genotyping for AD. The cardiologist might conclude that there is no clear and well-established risk of developing AD revealed by the test, and therefore no need to raise that possibility with the patient. If we consider more controversial associations for which genetic testing could be carried out, such as the alleged connection between certain genes and violent behavior, one can argue that the physician is justified in not informing the patient about such possible associations because they are too hypothetical. Physicians ought to avoid misleading their patients, the argument could go, and uncertain information can sometimes be more misleading than no information. There are two responses to this argument. First, there is no basis for the general claim that patients will be misled. Uncertain or ambiguous information is not necessarily misleading--presumably, the physician is not misled by it. Consequently, whether a particular patient would be unavoidably misled by it is an empirical matter. Second, physicians must bear in mind that they are not the only sources of information their patients have. Genetics is a favorite topic for many reporters, but not all the reporting in newspapers or television on advances in genetics is accurate or even responsible. Physicians have no control over what fills their silence, [End Page 136] and so cannot claim that being silent will protect patients from being misled. If the concern is about not misleading the patient, then surely the patient is less likely to be misled by a careful disclosure from a professional than by the haphazard gleanings of a report on the evening news. A third way of defending the partial disclosure approach is to invoke a simplified version of the doctrine of double effect. As this principle is understood in many medical discussions, an action with bad consequences can be morally justified as long as these consequences were unintended and the action was performed with a good intention. This doctrine is usually discussed in the context of treating those at the end of life. Although deliberately causing the death of a terminally ill patient is prohibited, administering high doses of pain medication to relieve the patient's suffering is permissible, according to the doctrine, even though this treatment may hasten the person's death (Quill, Dresser, and Brock 1997). Accordingly, one might argue that if assessing CAD risk is the intended use of APOE testing, the unintended but foreseen negative consequences of learning of an AD risk do not determine the ethical character of the test. The doctrine of double effect is extremely controversial, so appealing to it may not be a good strategy. But even if we put the objections to that doctrine aside, its applicability to the issue of pleiotropic tests is too strained to be useful. The doctrine is invoked in discussions regarding the permissibility of certain actions. However, the issue with pleiotropic testing is not so much the permissibility of testing as what should be acknowledged and disclosed about what the testing may reveal. Even staunch supporters of the doctrine of double effect do not claim that the unintended bad consequences of an action need not be acknowledged or disclosed. But the most serious problem with the partial disclosure approach is that it ignores (or violates) the need for pre-test informed consent. If we hold that patients have the right to refuse genetic tests because of what might be discovered, then consent is required in order to administer the tests. In the case of pleiotropic tests, if patients are not informed of the unintended as well as the intended uses of the tests, their consent is clearly not informed. Their right to refuse to be tested has not been respected. These difficulties with partial disclosure lend support to the practice of consolidation. This approach consists of informing the patient that the pleiotropic genetic test is, in fact, pleiotropic. The patient is told about the unsought information that the pleiotropic test might reveal and must consent to the unintended as well as the intended medical uses of the test [End Page 137] before it can be administered (cf. Post et al. 1997). The main argument for consolidation lies in its apparent emphatic acknowledgment that the patient's autonomy must be respected and that such respect requires not withholding information even if it pertains to an unintended use of the genetic test. One unfortunate consequence of this approach is that people who would benefit from a particular test--as in CAD risk assessment--might refuse because of worries about the unintended information obtained--as in AD risk assessment. But this consequence is of a piece with that resulting from a patient's refusal of treatment and thus is acceptable. A different problem with consolidation concerns the clinical requirements for each medical use of the test. When two separate tests are given, the clinical requirements for each can be quite different. However, as one of the few writers who has discussed pleiotropic genetic tests has argued, the clinical requirements regarding any use of a pleiotropic test would be set by its most problematic use (Juengst 1998). The reason for this is that the potential harms resulting from a test need not depend upon which use of the pleiotropic test is intended. Insofar as a concern about these harms shapes the clinical requirements for administering a test, they would seem to govern the test, regardless of its intended use. The most restrictive requirements--which would be those associated with the most problematic use of the test--would therefore dominate. The component of this requirement or "standard of practice" at issue here is genetic counseling. It has become common to assert the need for some counseling whenever genetic testing is performed or even considered, with the nature and amount of counseling depending on the condition at issue. From the standpoint of the consolidation approach, genetic counseling for a pleiotropic test would have to cover its unintended uses as well as its intended ones. Educating the patient about the various kinds of information the pleiotropic test could yield would not seem to be especially burdensome for the physician, who should know about the different clinical uses of any pleiotropic test he or she would propose to administer. But if the clinical requirements or the standard of practice for administering a test call for an extension of the clinician's responsibilities to include a great deal more, such as making available "psychiatric care, support group, and pastoral care" (cf. NIA/AA 1996, p. 1093), these duties might go beyond the resources, time, and abilities of physicians within particular specialties. A cardiologist might not have the appropriate professional expertise and resources that, according to the standard of practice, are essential for [End Page 138] adequate AD testing. Thus, Juengst (1998) reluctantly concludes that because the cardiologist generally cannot meet this component of the standard of practice, she will have to abandon APOE genotyping as a clinical tool. This argument is not about principles, but about contingent matters of current institutional structures and resource allocation. Hence, one could respond by arguing for appropriate changes in resources, and the like, to accommodate this need for genetic counseling for pleiotropic tests. For example, one could maintain that cardiologists administering the APOE test should be equipped with the resources to provide their patients with genetic counseling regarding AD risk. I would like to explore a different response to this argument. Many of the problems raised by pleiotropic tests involve the harms that can arise from possessing genetic information about oneself. I want to examine more carefully the concern with those harms and the belief that there is a routine need for genetic counseling in connection with genetic testing. I noted earlier that receiving genetic information is sometimes portrayed as a situation fraught with potential social and psychological harms and that genetic counseling is therefore considered necessary to help the patient cope. This is a distortion of the purpose of genetic counseling. The need for genetic counseling is not based on the special severity of the harms associated with genetic information; it is based on a reaction to the history of eugenics. In an effort to emphatically distance themselves from the uncomfortably recent history of eugenics, health professionals take special measures--typically by providing genetic counseling services--to help people make decisions about the use of genetic information without coercion or other threats to their autonomy (Wachbroit and Wasserman 1995). The expansion of these services to include efforts to help the patient anticipate and cope with the results of genetic testing, while not itself objectionable, has led to an assumption that the act of obtaining genetic information subjects the patient to otherwise avoidable social harms and psychological burdens. This act takes on the aspect of engaging in risky behavior, justified only if there are sufficient medical benefits. If we take the historical rationale seriously, the strongest case for genetic counseling is in the area of reproduction. This was where most eugenics policies were targeted and where we therefore need to be most sensitive to the possibilities of coercion or other threats to autonomy. The expansion of genetic counseling to other areas of medicine, however, where [End Page 139] there is no reason to expect a special threat to autonomy, encourages a "genetic exceptionalism," as if the mere fact that genetic methodologies were used requires providing the patient with special support and care. This genetic exceptionalism is dubious at best. Compare the case of a patient who learns he is at risk of heart disease from an echocardiogram with the case of a patient who learns that she is at risk of heart disease from a genetic test. It seems strange to claim that the latter patient, as a matter of routine procedure or of clinical requirements should receive counseling, but not the former. There need be no difference between these two cases regarding possible social or psychological harms. Of course, in certain situations and for certain individuals, counseling and other support services are indicated and appropriate. But to insist that counseling should be routinely provided whenever the medical test is genetic leads to a mystique surrounding the idea of genes that encourages misunderstandings about what genes are and distorts their significance. (It should be noted that one could reject genetic exceptionalism but take the argument in a different direction: All medical tests, genetic and nongenetic, should routinely be accompanied by pre- and post-test counseling. This would involve either a drastic strain on resources or a radical reallocation. The merits of either proposal deserve more discussion than I can give here.) If a patient, because of his psychological or emotional state, should be offered special counseling services before undergoing AD testing, then those services should also be in place if he undergoes CAD testing using APOE genotyping. But neither the genetic character of this test nor the information it may reveal requires that such services be routinely provided. Absent special concerns about the patient's psychological state, there is no need for the cardiologist to provide special clinical services such as pastoral or psychiatric care. The cardiologist, therefore, can use APOE testing as a clinical tool, informing the patient of the other clinical uses of the test. Indeed, even if the test were used for AD risk assessment, these extraordinary counseling services would not be routinely required procedures. The situation is different, however, if the use of the pleiotropic test is related to reproduction. Then Juengst's argument that any use is subject to the standards governing the most problematic use prevails and counseling services would be required. But in such cases, this will not be much of an added burden, since reproductive testing programs will (or should) already have these services in place. [End Page 140] Is There an Obligation to Know? Finally, it is time to challenge one of the assumptions underlying the recommendation that APOE genotyping for AD risk would be inappropriate. Recall that the basis for the recommendation is concern that uncertain or ambiguous test results could give rise to considerable social and psychological harms without offsetting medical benefit. This assessment, however, rests on construing harms too broadly and benefits too narrowly. In order to make the discussion of this point clearer, let us suppose that the uncertainty or ambiguity of the information were no longer a problem. Would the recommendation against direct AD testing be justified? A principal harm often cited in assessing the costs of AD testing is the risk of discrimination and stigmatization. This harm is real and, perhaps, substantial. Nevertheless, we should keep in mind where the fault lies. If a patient becomes uninsurable when a genetic condition is discovered, the proper target for criticism is the insurance industry, not the discovery. If someone suffers unfair employment discrimination because of a genetic test result, the fault lies with the practices of the workplace, not with the information revealed by the test. No thoughtful person should be misled on this point. Many discussions--not only in the ACMG recommendations, but also in arguments that have sometimes been made on behalf of a right not to know--are based on the assumption that the sensible response to potential harms arising from genetic information is to avoid obtaining the information, as if by not knowing or not obtaining the information the harm has been averted. The inadequacy of this reply can be made clear by imagining how this dilemma would look in a different context. Consider the harms associated with the prejudice and discrimination suffered by African-Americans. Although the victims of this discrimination are usually identified by skin color and other features, the structure of the prejudice often follows a "one drop of blood" rule: the offspring of African-Americans may be subject to the same discrimination even if they do not happen to have the skin color or features typically associated with AfricanAmericans. Imagine the case of someone whose parents are unknown and whose skin color and features are "White." It might be thought risky for such a person to trace her roots. Information about her parents might result in social harms and psychological burdens if it indicates that she has an African ancestry and she then suffers racial discrimination as a result. Nonetheless, most [End Page 141] people would agree that advising her not to seek information about her ancestry, or to do so only if counseling were available, would be offensive even if the advice were well-meant. It would presume that her legitimate interests in knowing about herself should defer to the costs arising from unjust social practices. Of course, I am not suggesting that one's race is like a disease; my point is that many of the social and psychological risks associated with the discovery of racial heritage and of genetic conditions are comparable and should receive no greater weight in one context than in the other. At the same time, the benefits of AD testing, assuming a reliable test, have been obscured by an overly narrow focus on medical outcomes. I certainly do not wish to minimize the harms of genetic discrimination and other burdens. Nevertheless, the prevailing picture is too one-sided. It leaves out the value we can sometimes discern in knowing our genetic condition even in the absence of medical benefits. (In the racial example above, citing possible medical benefits in knowing one's ancestry hardly captures the importance someone might attach to knowing.) It also fails to acknowledge the obligation we sometimes have to know our genetic condition. The idea that we can sometimes have an obligation to know our genetic condition has not received nearly as much attention as have the harms arising from such knowledge, but it is at least as significant. Most of us are enmeshed in a network of personal obligations and commitments to family members, dependents, and other loved ones. In many cases, if we have information about our medical condition, we can more effectively discharge these obligations, or at least avoid measures that, under the circumstances, may be futile. Consider the case of a 40year-old parent of minor children who refuses to know whether he is at high risk of contracting AD within the next ten years. His refusal might be irresponsible; it might amount to a failure to engage fully in the planning that is indicative of a parent's commitment to his children. The strength of the moral obligation to know one's genetic condition will depend upon the particulars of the situation. In all likelihood, however, a person's responsibility to know will not depend upon the strength of his or her desire to know or not to know. The idea of having an obligation to know can seem jarring at first. We are drawn to a picture of an individual, faced with the prospect of knowing, weighing how that knowledge would affect the quality of his or her [End Page 142] life. The thought that someone ought to know seems to go against our cultural assumptions, as if such an obligation were an unwelcome interference in the private relationship a person has with his or her life. The problem with this picture of solitary individuals contemplating whether to know about their future is that it fits so few of us. The assessment of a genetic test is therefore much more complicated than any weighing of medical benefits against social or psychological harms. Indeed, as the above example suggests, there can be an obligation to know that trumps the considerations of these harms and benefits. This trumping would cut across all uses of a pleiotropic test. Robert Wachbroit, Ph.D., is a Research Scholar at the Institute for Philosophy and Public Policy at the University of Maryland. References ACMG/ASHG. American College of Medical Genetics and American Society of Human Genetics Working Group on ApoE and Alzheimer Disease. 1995. Statement on Use of Apolipoprotein E Testing for Alzheimer Disease. Journal of the American Medical Association 274: 1627-29. Juengst, Eric T. 1998, in press. The Ethical Implications of Alzheimer Disease Risk Testing for Other Clinical Uses of APOE Genotyping. In Genetic Testing for Alzheimer Disease: Ethical and Clinical Issues, ed. Stephen G. Post and Peter J. Whitehouse. Baltimore, MD: Johns Hopkins University Press. Mayeux, Richard; Saunders, Ann M.; Shea, Steven; et al. 1998. Utility of the Apolipoprotein E Genotype in the Diagnosis of Alzheimer's Disease. New England Journal of Medicine 338: 506-11. NIA/AA. National Institute on Aging and Alzheimer's Association Working Group. 1996. Apolipoprotein E Genotyping in Alzheimer's Disease. Lancet 347: 1091-95. Post, Stephen G.; Whitehouse, Peter J.; Binstock, Robert H.; et al. 1997. The Clinical Introduction of Genetic Testing for Alzheimer Disease: An Ethical Perspective. Journal of the American Medical Association 277: 832-36. Quill, Timothy E.; Dresser, Rebecca J.; and Brock, Dan W. 1997. The Rule of Double Effect--A Critique of Its Role in End-of-Life Decision Making. New England Journal of Medicine 337: 1768-71. Strittmatter, W. J.; Saunders, A. M.; Schmechel, D.; et al. 1993. Apolipoprotein E: High Avidity Binding to Beta-Amyloid and Increased Frequency of Type 4 Allele in Late-Onset Familial Alzheimer Disease. Proceedings of the National Academy of Sciences 90: 1977-81. Wachbroit, Robert. 1996. Disowning Knowledge: Issues in Genetic Testing. Report from the Institute for Philosophy and Public Policy 16: 14-18. ------, and Wasserman, David. 1995. Patient Autonomy and Value-Neutrality in Nondirective Genetic Counseling. Stanford Law & Policy Review 6: 103-12. Wilson, P. W. F.; Myers, R. H.; Larson, M. G.; et al. 1994. Apolipoprotein E Alleles, Dyslipidemia, and Coronary Heart Disease. Journal of the American Medical Association 272: 1666-71. http://muse.jhu.edu/journals/kennedy_institute_of_ethics_journal/v008/8.2wachbr oit.html.